TechPrufer
Get started
PlanetScale
PlanetScale logoPlanetScale · Feature trail

Postgres Full-Text Search (TIN)

5 posts · Jul 2022 – Oct 2026

What this trail covers

PlanetScale offers TIN, a high-performance full-text search extension for PostgreSQL. This post details updates including stemming support for 18 languages, improved tiebreaker handling in top-k queries, and optimizations that restore TIN's performance lead over competitors like ParadeDB, particularly in concurrent write scenarios. The benchmarks highlight TIN's speed advantage on both x86-64 and ARM64 architectures, with specific performance metrics provided for conjunction, disjunction, and phrase queries. The post also discusses the trade-off of omitting common terms from BM25 scoring by default and how to tune this behavior.

All PlanetScale trails
  1. TIN v1.0.6: 2-3x faster, with support for stemmingThis post details the release of TIN v1.0.4 and v1.0.6, introducing stemming support for 18 languages using the rust_stemmers crate and Snowball project algorithms. It also describes optimizations to the tiebreaker logic for top-k queries, ensuring consistent performance with equal scores. Benchmarks are presented comparing TIN against ParadeDB on conjunction, disjunction, and phrase queries across x86-64 and ARM64 architectures, demonstrating TIN's regained performance lead, especially in concurrent write scenarios. The post also elaborates on the BM25 scoring optimization of omitting common terms and how to adjust this behavior.
  2. TIN Postgres search is faster, better, and cheaperThis post demonstrates how TIN's performance advantages enable enhanced search features like immediate search on keystroke (no debouncing), autocomplete via wildcard matching, fuzzy matching using Levenshtein distance, and indexing of stop words. It highlights TIN's ability to provide exact counts for large result sets quickly, contrasting it with slower, less accurate alternatives. The post also details cost savings compared to other solutions.
  3. Anatomy of a (Postgres) search engineThis post provides a deep dive into the technical implementation of full-text search, specifically focusing on the inverted index data structure. It explains the components of an inverted index (term dictionary, postings lists, positional and frequency data), the process of tokenization (including stemming and stop words), and scoring mechanisms like BM25 for top-k queries. It also touches on index size trade-offs and the management of segments (mutable vs. immutable) for dynamic data. This article serves as a foundational explanation of how full-text search works, complementing the existing feature thread on PlanetScale's TIN extension.
  4. Introducing TIN: full-text search for PostgresThis post introduces TIN, a new full-text search extension for PostgreSQL. It details the design goals, including support for boolean expressions, phrase queries, fuzzy matching, and ranked results. Extensive benchmarks are presented comparing TIN against ParadeDB, pg_textsearch, and Postgres GIN across various workloads (conjunction, disjunction, phrase, count queries) with and without concurrent writes. The benchmarks highlight TIN's superior QPS, lower p99 latencies, and reduced MB/query compared to alternatives, demonstrating its performance advantages in build time, index size, and query execution.
  5. How do Database Indexes Work?This post explains the fundamental concepts of database indexes, focusing on their role in speeding up read queries. It details how indexes, particularly B-tree structures, reduce query complexity from O(N) to O(log N) by creating ordered data structures with pointers to records. The post covers different index types (keys, unique indexes, text indexes) and their characteristics, including the use of `FULLTEXT` indexes with `MATCH()...AGAINST()` in MySQL. It also touches on index-only scans and the trade-offs of having too many indexes, such as increased storage space and slower INSERT operations. The `EXPLAIN` statement is highlighted as a tool for verifying index usage. The post also provides the syntax for creating indexes in MySQL and discusses various index options and types.